INDEX

Explanations

mathematical formulas and code

The neuron responds to numeric tokens—especially multi‐digit numbers and decimal values—i.e. it picks out numeric data in the text.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

<bos>

-1.75

 También

-1.55

ández

-1.40

⌒

-1.38

 juiz

-1.37

dő

-1.34

edonia

-1.34

⾔

-1.30

 élastique

-1.30

 Meksi

-1.30

POSITIVE LOGITS

1.46

ate

1.45

↵

1.44

êmes

1.34

1.32

on

1.27

 still

1.25

 fiecare

1.25

1.24

1.21

Activations Density 0.163%